Prev: upgrade RAM vs CPU
Next: Parallel Computing; File Dependencies
From: Wayne King on 9 Aug 2010 09:53
"Debora " <debi_can(a)yahoo.com.br> wrote in message <i3otib$rjd$1(a)fred.mathworks.com>...
> Sorry, let me try to explain further.
>
> First, yes, I have used the filter design toolbox for the butterworth filter.
>
> As for the signal, I may not be using the appropriate terms, so I will try to write about it. My data has frequencies up to 1000 Hz. I am considering, from literature and previous works, that emg frequencies are from 20 to 500 Hz. Those are the frequencies I've used. Under 20 Hz, close to 0 Hz, I have this peak. This always happens with the equipment I've used, so I am sure this is noise and must be removed.
>
> Thanks, again,
> Debora

Hi Debora, Assuming that you really want to remove this peak (I'm speaking about Jan's post). Then:
1.) Identify the frequency corresponding to the peak.
2.) Use iirnotch() from the Filter Design Toolbox to remove the oscillation.

For example, just creating some data here:

Fs = 1000;
t = linspace(0,1,1000);
% just simulating a 50-Hz component
x = cos(2*pi*50*t)+randn(size(t));
% identifying the frequency
h = spectrum.periodogram;
msspec = msspectrum(h,x,'Fs',1000,'NFFT',length(x));
[y,index] = max(msspec.Data);
freq = msspec.Frequencies(index);
% convert that frequency to normalized frequency
W = (2*freq)/Fs;
BW = W/35;
% get the notch filter
[b,a] = iirnotch(W,BW);
y = filtfilt(b,a,x);
msfilt = msspectrum(h,y,'Fs',1000,'NFFT',length(x));
subplot(211);
plot(msspec); title('Before Notch Filtering');
subplot(212);
plot(msfilt); title('After Notch Filtering');

Hope that helps,
Wayne
 | 
Pages: 1
Prev: upgrade RAM vs CPU
Next: Parallel Computing; File Dependencies